1242 Inorganic Chemistry, Vol. 10, No. 6,1971
C. S.SPRINGER, JR., R. E. SIEVERS,
AND
B. FEIBUSH
CONTRIBUTION FROM THE DEPARTMENT OF CHEMISTRY, STATE UNIVERSITY OF REWYORKAT STONY BROOK,STONY BROOK, NEW YORK 11790, AND THE -4EROSPACE RESEARCH LABORATORIES, ARC, WRIGHT-PATTERSON AIR FORCE BASE,OHIO 45433
Chiral Chelates with Chiral Ligands. The Stereoisomers of Tris [( )-3-acet ylcamphorato] cobalt (111)
+
BY CHARLES S. SPRINGER, J R . , * ~ROBERT E. SIEVERS,
AND
BINYAMIN FEIBUSH
Recived October 23, 1970 T h e equilibrium mixture of the four diastereoisomers of tris-[( +)-3-acetylcamphorato]cobalt(III) a t 60’ in CeH6 has been found to contain 45.2 mol % A-trans, 31.3 mol % A-trans, 16.4 mol yo A-cis, and 7.0 mol % A-cis isomer. These values indicate that, contrary to previous statements in the literature, the 3-acylcamphorato ligand exhibits only slight thermodynamic stereoselectivity in tris complexes. T h e four isomers have been separated and characterized individually by ’H nmr, ORD, and CD. A study of their CD curves reveals that a significant vicinal effect is extant in the visible region of the spectrum and that it is roughly additive to the configurational effect.
The two main factors which historically have been of interest in the study of chiral chelates with chiral ligands are (1) the extent of stereo~electivity~ exhibited by the chelate and ( 2 ) the contribution to the optical activity of the metal transitions made by the presence of chirality in the ligand, the vicinal e f f e ~ t . ~ The majority of bidentate chiral ligands which have been employed in such studies are lj2-diamines or a-amino acid anions. Another type of bidentate chiral chelating ligand has been intimately connected with the development of such studies since their very beginnings. This is the P-keto enolate produced by the acylation of camphor in the 3 position (the (+) form is shown).
of only one isomer each of the Co(II1) and Cr(II1) tris complexes using the (+)hmc ligand. These were (-)Co((+)hmc)s and ( + ) C r ( ( + ) h r n ~ ) , . ~If~ only one chiral form of the ligand is used, there are four possible diastereomers, for example, A-trans-M (( +)hmc)a, A-cisM((+)hmc)a, A-trans-M((+)hm~)~, and A-cis-M((+)hmc),. Since Lifschitz observed no Cotton effect in the visible region of the spectra of compounds such as Co((+)hmc)z .2H20 and Ni((+)hmc)z.2H20,7b he concluded that any ligand such as this, in which the asymmetric carbons are not contained in the chelate ring, is incapable of causing a vicinal effect. Thus, he reasoned that the Cotton effect observed in the visible region of the spectrum of C ~ ( ( + ) h m c ) indicated ~~” a tetrahedral s t r ~ c t u r e . 7 ~In 1938, Pfeiffer and his coworkers observed no Cotton effect in the visible spectrum of Ni(( + ) h m ~ ) and, ~ following Lifschitz’s reasoning, assumed i t to have a planar structure.s French and Corbett disagreed, on the basis of the observed paraThe formyl derivative (R = H), abbreviated in the showed that magnetism and uv Cotton e f f e ~ t . Mellor ~ literature as oca, for oxymethylenecamphor, or hmc, the Cotton effect was due to a ligand transition and that for hydroxymethylenecamphor, was first prepared and both Ni( (+)hmc)z and the bis(benzoy1camphor)nickelused as a ligand in 1894.5 The spectra and mutarota(11) complex exhibited visible spectra typical of octations of the complexes of a number of metals with the hedral nickel(I1) , l o He concluded that vicinal effects benzoyl (R = CeHb) derivative were reported by Lowry could not be operative when the metal ligand bonds and coworkers.6 Although they claimed the detection were “ionic.” Lifschitz reinvestigated these systems of stereoisomers of the bis (benzoylcamphor)beryllium(II)6b,e and tris(benzoylcamphor)aluminum(III)6c~f in 1950 and reported the partial isolation of a (-)Cr((+)hrnc)a isomer.’l He could observe no Cotton effect complexes, the first systematic investigations into the in the visible region of a “racemic” mixture of the (+) stereoselectivity caused by these bulky chiral ligands and (-) isomers and felt that this confirmed his earlier were begun by Lifschitz.’ He reported the preparation conclusion regarding the lack of a vicinal effect. (1) Presented, in part, a t t h e 157th National Meeting of the American Chemical Society, Minneapolis, Minn., April 16, 1969. Mason and his coworkers used circular dichroism to (2) Author t o whom correspondence should be addressed a t t h e S t a t e assigned the A 1 2 configuration to the (-)Co((+)hmc)s University of New York a t Stony Brook. and (+)Cr((+)hmc), c o m p l e x e ~ . ~Dunlop, ~ Gillard, (3) We use t h e t e r m “stereoselectivity” in the same sense a s t h a t of Dunlop and Gillard, who defined i t as “the behavior of molecular diastereoand Ugo reported the separation of the (-) cobalt isomers:” J. H. Dunlop and R . D. Gillard, Advan. Iizorg. Chem. Radiochem., species into cis and trans isomers in 1966.14 They were 9, 185 (1966). (4) T h e t e r m “vicinal effect” was first used for metal complexes by Y . Shimura, Bull. Chem. Soc. J a b . , 31, 315 (1958). See also J. Fujita a n d Y. Shimura, “Spectroscopy and Structure of Metal Chelate Compounds,” K . Nakamoto a n d P. J. McCarthy, E d . , Wiley, N e w York, h‘. Y.,1968, Chapter 3 , p 156. (5) A . W . Bishop, L. Claisen, and W. Sinclair, .An%, 281, 314 (1894). (6) (a) H. S. French a n d T. M. Lowry, Proc. Roy. SOL.,Sev. A , 106, 489 (1924); (b) H. Burgess and T. ?A, Lowry. J. Chem. Soc., Loizdoiz, T V Q M S . , 125, 2081 (1924); (c) I . J. Faulkner a n d T . AM. Lowry, ibid., 127, 1080 (1925); (d) T. M. Lowry, H. Burgess, I. J. Faulkner, and R. C. Traill, Proc. Roy. Soc., Seu. A , 132, 387 (1931); (e) T. M. Lowry and R . C. Traill, ibid., Sei,. A , 132, 398 (1931); (f) T. &I. Lowry and R . C. Traill, ibid., S P Y . A , 132, 416 (1931). (7) (a) J. Lifschitz, Red. TYQU.Chim. P Q Y S - B Q S41, , 627 (1922); (b) J. Lifschitz, Z . P h y s . Chem., 106, 27 (1923); (c) J. Lifschitz, i b i d . , 114, 485 (1925).
(8) P. Pfeiffer, W. Christeleit, T . Hesse, H. Pfitzner, a n d H. Thielert, J . Prakt. Chem., 160, 261 (1938). (9) H. S. French a n d G. Corbett, J . Amev. Chem. Sac., 62, 3219 (1940). (IO) D. P. Mellor, J . Proc. R o y . Sac. N . S. W . , 76, 157 (1942). (11) J. Lifschitz, Recl. Trav. Chim. P Q Y S - B U69, S , 1495 (1950). (12) T h e symbols A and A , as used in this paper, are consistent with t h e recently published nomenclature proposals: Inorg. Chem., 9, 1 (1970). Thus, A signifies t h e absolute configuration in which t h e chelate rings have a left-handed helical relationship when viewed along the Ca or pseudo-Cg axis of a tris-bidentate chelate. (13) (a) R. E. Ballard, A. J. McCaffery, a n d S. F. Mason, Proc. Chem. Soc., London, 331 (1962); (b) S. F. Mason, Quavl. Rev., Chcm. Sac., 17, 20 (1963); ( c ) A. J. McCaffery, S. F. Mason, and R . E. Ballard, J . Chem. Soc., 2883 (1965). T h e cobalt complex is incorrectly designated (+)Co((+)hrnc)s in this paper. (14) J. H. Dunlop, R. D. Gillard, and R. Ugo, ibid., A , 1540 (1966).
TRIS [(+I
l
Inorganic Chemistry, Vol. 10,No. 6,1971 1243
-3-ACETYLCAMPHORATO ]COBALT (111)
unable to isolate Lifschitz's (-) chromium isomer but assigned the (+) isomer the trans structure and reported the isolation of the A-trans and A-trans isomers of R h ( ( + ) h m ~ ) ~thus , affording the first example of a A configuration in the (+) -3-acylcamphor series. They concluded that stereoselectivity in tris complexes of the (+) forms of the ligands produced the order of stability A-trans > A-cis > A-trans > A-cis and that A isomers could only be isolated in systems under kinetic control. Chen and Everett reported an investigation of the complexes of Co(II1) and V(II1) with (+)- and (-)hmc and (+)atc, the acetyl derivative (R = CH,).16 Their results with (+)hmc agreed with those of Dunlop, Gillard, and Ugo and they reported that the two (+) cobalt isomers which they detected with the (-)hmc ligand had the expected A-trans and A-cis structures; ie., the (-) form of the ligand stabilizes the A configuration. Their results with (+)atc were entirely analogous to the (+)hmc findings. They detected no evidence for the existence of the A-trans or A-cis isomers of the cobalt complexes with the (+) ligands. However, their nmr studies of the labile paramagnetic V(II1) chelates indicated that the Atrans, A-cis, and A-trans isomers of the V ( ( + ) a t ~ ) ~ complex did coexist in equilibrium and that all four isomers of the V((+)hmc)a complex could be detected. Lintvedt and Fatta have reported infrared spectral and molecular weight investigations of a large number of hmc complexes and have shown that the Ni(hmc)n and Co(hmc)n complexes are trimeric with octahedral coordination and that the Cu(hmc)z complex has a planar structure.I6 We have accomplished the preparation, isolation, and characterization of all four diastereomers of the C ~ ( ( + ) a t c )complex. ~ In so doing, we have discovered that, contrary to the almost complete agreement in all of the relevant literature cited above, the tris chelates of the 3-acylcamphorato ligands show only slight stereoselectivity and do indeed exhibit a significant vicinal effect. l 7 Experimental Section
+
Preparation of the Ligand.-( )-3-Acetylcamphor was prepared from (+)-3-bromocamphor'* and ethyl acetate according to the organozinc syntheses of B r ~ h 1 . l ~ The product is a colorless liquid a t room temperature and was purified by triple distillation under reduced pressure, the last distillation being under a nitrogen atmosphere: bp 74" (0.45 mm), 76" (0.60 mm), 79" (0.70 mm), 81-83' (0.80 mm), 86-89' (1.30 mm) [lit.I9 115-117" (10 mm), 118-118.5' (11.2 mm), 127-128' (15.5 m m ) ] . The neat ligand possessed an ir spectrum generally similar to that reported for solid H(hmc).l6 Some of the details of this spectrum as well as those of the nrnr spectrum will be discussed in the Results. The ORD curve of the ligand was also recorded; [ o ~ ] f50.2' ~~D (CHaOH). Preparation of the Complex.-Tris [ (+)-3-acetylcamphorato]cobalt(II1) was prepared from Naa[Co(COs)s].3HzO and H((4- )atc) according to the method of Gillard.I4 After thereaction was completed, the reaction mixture was filtered to remove the (15) Y .T. Chen and G. W. Everett, J . Ameu. Chem. Soc., 90, 6660 (1968). (16) R. L. Lintvedt a n d A. M. F a t t a , I n o v g . Chem., 7 , 2489 (1968). (17) K i n g a n d Everett have recently reinvestigated the C o ( ( + ) a t c ) 8 system a n d have succeeded in separating all four isomers by thin layer chromatography. Their isomeric abundance values are in good agreement with ours but their conclusions concerning the degree a n d source of stereoselectivity differ somewhat: R. M. King a n d G . W. Everett, Jr.; i b i d . , 10, 1237 (1971). (18) T h e absolute configuration of (+)-3-bromocamphor has been determined by anomalous dispersion X-ray diffraction techniques: F. H. Allen a n d D. Rogers, Chem. Commun., 837 (1966). (19) J. W. Bruhl, Chem. Bev., 8 1 , 746 (1904).
brown, gummy side product. The upper deep green benzene layer of the filtrate contained the complex and was separated from the lower colorless aqueous layer. The benzene layer was then rapidly passed through a very small column of acid-washed alumina to remove any soluble Co(I1) impurities. The nmr of the concentrated benzene layer was studied to determine the relative proportions of the stereoisomers. (See the Results.) The relative areas of the various peaks were measured by the weighing of paper cutouts or by planimetry. Chromatography of the Complex.-A typical separation and purification of a reaction mixture, treated as stated above, will be given. For a reaction mixture containing approximately 7 g of complex in approximately 25 ml of benzene, a chromatographic column of 4-cm diameter was tightly packed under benzene with approximately 500 g of acid-washed alumina (Merck No. 71695). Benzene was allowed to run out until the liquid level was just a t the top of the column (height approximately 40 cm). The reaction mixture was added to the column and benzene was run out from time to time to keep the liquid level just a t the top of the column. As soon as all of the complex was absorbed a t the top of the alumina, the rest of the column was filled with freshly distilled benzene to produce a head approximately 115 cm above the bottom of the column. Maintaining this head, a flow rate of approximately 10 ml/min was obtained. As the elution proceeded, the initial green band separated into two new bands. The faster moving band was approximately 3 times larger than the more strongly absorbed band. The leading portion of the larger band developed a distinct brownish green tint while the trailing portion became deep green. The opposite was noted for the smaller band; i.e., the trailing portion had the brownish green tint. After approximately 0.5 hr, t h e breakthrough of the leading band occurred. I t was collected in two approximately equal fractions as was the second band which broke through after approximately 8.5 hr. The latter band was completely eluted after approximately 11hr. The isomeric contents of the fractions could be monitored by nrnr spectroscopy. Each band was found to contain two isomers. The two isomers in the larger band were labeled 01 and p in the order in which they came off the column; is.,P was the more strongly absorbed isomer. Similarly, the y isomer was eluted before the 6 isomer in the second band. A second chromatographing of each fraction on a smaller column was sufficient to allow collection of reasonable amounts of the pure isomers. The single band produced by each fraction was itself collected in two fractions of approximate relative ratios of 3 : 1 or 1:3, as the case required. Thus, the first fraction of the a/3 band was collected in two fractions, the first one of which comprised three-fourths of the total. I t contained pure 0 1 . I n the same way, pure samples of the p, y, and 6 isomers were also collected. Quantitative separation was not achieved and, therefore, measurement of the relative amounts of the various isomers was achieved by integrating the nmr spectrum of the reaction mixture as described above. Characterization of the Isomers.-The four isomers melted with decomposition in sealed, evacuated tubes a t the following temperatures: 01, 196-198"; p, 200-202'; y, 150-154" (did not decompose until 187'); 6, 193-196" (all temperatures are uncorrected). Elemental microanalyses were conducted by Galbraith Laboratories and the results are presented in Table I. TABLE I ELEMENTAL ANALYSESFOR THE ISOMERS OF Co(( +)atc)a % C
% H
Calcd Found
67.69
8.04
%co 9.22
a
67.96 67.06 67.03 67,22
7.95 7.96 7.85 8.17
8.98 9.62 8.26 8.72
P Y
6
The infrared spectra of the four isomers (as recorded in Nujol mulls on a Perkin-Elmer Model 137 instrument) were nearly identical and were in agreement with the spectrum reported for Co(hmc)3.16 For example, the C-0 and C-C stretching frequencies of Co(hmc)g were reported a t 1615 and 1481 cm-', respectively, and the analogous assignments for the Co((+ )atc)g isomers are as follows: 01, 1605, 1485 cm-'; p, 1610, 1495 cm-l; y , 1625, 1510 cm-l; 6, 1625, 1505 cm-l. Nuclear Magnetic Resonance Spectra.-The lH nmr spectra of the isomers were recorded on Varian A-60 or HR-100 spectrom-
1244
C. S.SPRINGER, JR.] R. E. SIEVERS, AND B. FEIEUSH
Inorganic Chemistry,Vol. 10,No.6,1971
TABLE I1 'H NMRSPECTRAL DATAFOR H((+)atc) 7
8b
Ex0 7 J , Hz Re1 area
IN
Endo J , Hz
,-bb
CClp 7
Re1 area
-------Anti-ab
J . Hz
Re1 area
2.80 0 0.85 3.24 4.4, 1.2 1.2 * , . ... ... 2.68 3.9 0.85 2.47" c c 2.74c c c 1.87 0 3.2 2.20 0 2.9 2.25 0 2.8 0.95 0 2.7 1.00 0 3.0 0.88 0 6.01 0.90 0 6.0d 0.82 0 6.08 0.58 0 2.6 I n ppm, downfield from TMS. The 4-H multiplets a Concentration approximately 2O%, by volume; temperature ambient. for the endo and anti isomers could not be resolved, The combined peak is a doublet with a splitting of 3.6 Hz. The higher field component is somewhat more intense than the low-field member. The 4-H protons are spin coupled with the exo 5 protons with a coupling constant of approximately 3.9 Hz. I n addition, the 4-H proton in the endo isomer is spin coupled with the exo 3 proton. Arbitrary area standard for those peaks assigned to the exo isomer. e Arbitrary area standard for those peaks assigned to the endo isomer. Arbitrary area standard for those peaks assigned to the anti isomer.
3-H 4-H Acetyl-olefin methyl 8-, 9-, and 10-methyl
eters and will be discussed in detail in the Results. All chemical shifts reported are the average values from several spectra and were measured by the side-band technique. Visible-Region Electronic Absorption, Optical Rotatory Dispersion, and Circular Dichroism Spectra.-The electronic absorption spectra of the four isomers in benzene were obtained on a Cary Model 14 recording spectrophotometer from 650 to 300 nm. The benzene solution ORD spectra were obtained on a Cary Model 60 recording spectropolarimeter over the same wavelength range. Beer's law compliance was noted from the fact that the straight line through the two values of cy500 for two different concentrations of the a isomer also passed through the origin of a plot of 01~03vs. concentration. The C D spectra were also obtained on the Cary 60. Thermal Equilibration of Isomers.--il solution of 21.8 mg of a-Co((+)atc)3 in 488.8 mg of Spectrograde benzene was degassed and sealed in a n nmr tube in vacuo. The solution was heated in a thermostated bath a t 60' until the nrnr spectrum, which was monitored on the thermally quenched sample a t intervals, showed no further change (after approximately 60 hr). No evidence of decomposition was noted. The nmr spectrum of the resultant mixture was analyzed for the relative amounts of the isomers in the same manner as was the reaction mixture.
Results The ligand, (+)-3-acetylcamphor, can exist in three tautomeric forms: a @-ketoenol form which has two diastereomersz0
syn
anti
a @-diketone form which has two epimeric diastereomers
v
/ A CH3 exo
C '0
C ' H3 endo
and another 6-keto enol form
(20) T h e nomenclature used here is analogous t o that used by Daniel and Pavia for H(hmc): A. Daniel and A. A. Pavia, Tetraheron Letl., 1145 (1967).
The H(hmc) analog of this last tautomer has been calculated to be of high energy on the basis of strain effects. The infrared spectrum of neat H((+)atc) shows four strong absorptions in the carbonyl-olefin stretching region a t 1750, 1710, 1695, and 1635 cm-1. The band a t 1750 cm-l can reasonably be assigned to the 2-carbony1 stretches of the exo and endo isomers which would not be expected to be very different from that of camphor (1750 cm-I 1 6 ) . The band a t 1710 cm-1 is most likely the carbonyl stretch of the anti isomer analogous to that of anti-H(hmc) (1711 cm-l) while the band a t 1635 cm-l is the accompanying C=C stretch [H(hmc) analog, 1628 cm-l].16 The relevant lH nmr spectral data and assignments for H((+)atc) in CC1, are given in Table 11. The spectrum is closely analogous, in the 3- and 4-hydrogen region (2.3-3.4 ppm downfield from TMS), to that reported for H(hmc) in the same solvent.20 Thus, the assignments in this region are made by analogy. The 4 protons are spin coupled to the exo 5 protons.23 The 3 proton (exo) in the endo isomer is spin coupled with the 4 proton (4.4 Hz) and with the exo 5 proton (1.2 Hz) which is related to it by a "W" c o n f i g u r a t i ~ n . ~The ~ relative amount of the anti isomer was obtained by subtracting an amount equal to the area under the 3-H peak of the endo isomer (-3.24 ppm) from the area under the combined 4-H peaks of the endo and anti isomers (-2.47 ppm). The relative amount of the exo isomer was obtained from the area under the 3-H peak a t -2.80 ppm. The mole percentages of the isomers thus obtained are as follows: exo, 40%; anti, 33%; endo, 279.',. The assignments in the acetyl-olefin and 8-, 9-, and 10-methyl regions are made from relative area considerations based on the various 3- and 4-hydrogen peaks. Three peaks present in the acetyl-olefin methyl region of the spectrum of the ligand indicate that essentially the same isomeric content also obtains in benzene solution. A well-resolved nmr spectrum of the neat ligand could not be obtained because of its high viscosity. Thus, its isomer content is not known. The relative amounts of syn and anti isomers of H(hrnc) have been reported to be solvent dependent.20-22 Our results are in agreement with those of Lintvedt and Fatta, who, in reinvestigating the low-field spectral region previously studied by Garbisch,21, 2 2 found no syn-H(hmc).16 It is interesting to note that the ratio z1p22
(21) (22) (23) (24)
E. W. Garbisch, J . Rmev. Chem. SOL.,86, 1696 (1963). E. W. Garbisch, ibid., 87, 505 (1965). J. I. Musher, M o l . P h y s . , 6 , 93 (1963), and references cited therein. K . M. Baker and B. R . Davis, Telvaheduon, 24, 1663 (1968).
Inorganic Chemistry,VoZ. 10,No. 6,1971 1245
TRIS[ (+)-~3-ACETYLCAMPHORATO]COBALT(III) TABLE I11 lH NMRSPECTRAL DATAFOR
,-SIsomer a
I
CeHe65.2" (6.0)F 42.6 (3.1)
ISOMERSOF C ~ ( ( + ) a t c ) ~
7CCl4---. 62.3 (2), 59.3 (l), 55.3 (l), 53.3 (2), 50.3 (3) 57.6 (2), 55.8 (l), 49.1 (3), 46.3 (3)
Acetyl methyl region
~ccl4---
C e H e 120.2 ( l ) , 116.3 (l), 114.6 (1)
of the concentrations of the exo isomer to the endo isomer reported by Daniel and Pavia for H(hmc) in CC14, 1.7, and in a CCt-DMSO mixture (where the syn isomer is favored over the anti isomer), 1.6,20is essentially the same for H((+)atc) in CC4, 1.5. Thus the major difference caused by the substitution of a methyl group for a proton in the R position of 3-acylcamphor seems to be a relative destabilization of the anti isomer. The four possible stereoisomers of a M((+)atc)a complex are shown in Figure 1. Since only one chiral
Figure 1.-The
Co(+atc)3 i n acetyl methyl region
-
124.8 ( l ) , 123.3 (2)
65.4 (1.3), 63.1 ( l . O ) , 119.7, 119.1 (2.0), 117.1 (1.3) 58.4 (1.5), 56.0 (2,1), 51.7 (1.2),44.3, 42.4, 41.0 (3.0) 58.0 (3), 50.1 (3), 118.3 (3) 61.6 (3), 54.2 (3), Y 47.3 (3) 41.7 (3) 59.7 (3), 54.6 (3), 67.5 (3), 64.3 (3), 118.9 (3) 6 42.7 (3) 50.1 (3) Chemice 3hifts are given in hertz downfic 1 from TMS. Relative areas are given in parentheses.
P
a
-
THE
9-, and 10-methyl region
127.3, 126.9 (2.0), 121.3 (0.9) 123.5 (3) 124.3 (3)
C6H6
8-,9-,10- " c a m p h o r " methyl region
stereoisomers of an M ( (+)atc)a complex.
form of the ligand is used, each isomer bears a diastereomeric relationship with the others. The two cis isomers have Ca symmetry while the two trans isomers have only CI symmetry. The three ligands in a cis complex are related to one another by the C3 axis and are thus equivalent. Therefore, only one nmr signal is to be expected for each of the different protons in a ligand. For a cis isomer there should be one acetyl, one 8-, one 0-, and one 10-methyl resonance. For a trans isomer, in which the ligands are diastereotopic, there should be three acetyl, three 8-, three 9-, and three 10-methyl resonances. Thus, a mixture of all four isomers could show as many as 8 lines in the acetyl methyl region (-1.9 to -2.1 ppm) and 24 lines in the 8-, 9-, and 10-methyl region (-0.7 to -1.2 ppm). When an nrnr spectrum of the benzene phase of the synthetic mixture is examined on an expanded scale, it is evident that more than one stereoisomer is present. As the various fractions were collected from the alumina chromatographic column, their nmr spectra were examined in the acetyl and 8-, 9-, and 10-methyl regions on expanded scale. Figure $ shows the spectra
I1
-2.0
-1.9
I
-1.1
-1.0
-0.9 -0.8
-0.7
8 (ppm)
Figure 2.-The 1H nmr spectra of the isomers of Co((+)atc)a in the acetyl and 8-, 9-, and 10-(camphor)methyl regions. The spectra were obtained in benzene a t ambient temperature. The relative areas are represepted by the number of vertical lines under the peaks (each line corresponds to one methyl group). The spectrum in the acetyl methyl region and that in the camphor methyl region are not necessarily shown with the same vertical scale for a given isomer.
of the four isomers, separated in this way, in these regions. Pertinent data are summarized in Table 111. The acetyl methyl region clearly shows that the Q and @ isomers are trans since they each exhibit three peaks. The y and 6 isomers are therefore cis. The 8-, 9-, and 10-methyl resonances are well resolved in each of these latter isomers. All but two of these resonances are resolved in the p isomer. In the Q isomer, there is a great
c.S. SPRINGER, JR., R. E. SIEVERS, AND B. FEIBUSH
1246 Inorganic Chemistry, VoL 10, No. 6,1971
deal of accidental degeneracy in the 8-, 9-, and 10methyl regions. It seems reasonable to suggest here that the cis isomers y and 6 are more strongly absorbed on the column because their molecular dipole moments are most likely greater than those of the trans isomers. This chromatographic behavior has been noted before in the separation of geometrical isomers of tris(Pketo enolato)cobalt(III) complexes.25 Recently, however, Everett and Chen have reported the following order of elution on a silica gel thin layer plate of the isomers of the Co(II1) tris complex of (-)-hydroxymethylenecarvone, a chiral P-keto enolate ligand : A-cis > A-trans > A-trans > A-cis (i.e., A-cis is the least strongly absorbed). 26 The electronic absorption spectra of the four isomers are almost identical. The energies and maximum extinction coefficients of peaks assigned to the ‘AI, ‘TI, (in idealized Oh symmetry) “d-d” transition are as follows: 16,200 cm-I (145), 16,300 cm-l (161), 16,200 cm-l (137); 16,300 cm-’ (235). The IA1, + lTag transitions were not clearly resolved but the circular dichroism studies (vide infra) show them to be centered a t about 22,500 cm-l. The weak intensities of these “d-d” transitions indicate that octahedral selection rules are in effect to a good approximation. Peaks a t -340 nm which are shoulders on more intense absorptions in the uv spectrum have intensities similar t o that of the Co(acac)3 transition a t 324 nm which recent SCF-MO calculations have indicated to be a metal d.rr + ligand X* charge-transfer t r a n s i t i ~ n . ~ ’Their energies and intensities (emax) are as follows: CY,29,500 cm-l (87 X l o 2 ) ;0,29,300 cm-‘ (87 X l o 2 ) ; y, 29,200 cm-l (70 X l o 2 ) ; 6, 29,200 cm-’ (70 X lo2). These spectra are in agreement with that portion of the spectrum of the trans fraction of Co((+)atc), shown by Chen and Everett.Is The cz and 6 isomers have a distinctly brownish green color when compared to the deep green of the 0 and y isomers. This is due to a shift of the minimum in the visible region from -545 nm for the p and y isomers to -555 nm for the CY and 6 isomers. This explains the band shading noted during the chromatography of the isomers The ORD curves for the four isomers are shown in Figure 3 The CY and 6 isomers are seen to be (+) iso---f
Cal+olc13
effects for the charge-transfer bands are exactly opposite those a t -615 nm for all the isomers. The signs of the Cotton effects of the bands a t -445 nm cannot be easily determined from the ORD curves (see, however, the CD spectra below). The maximum specific rotation of the P isomer in the “d-d” spectral region (-6750 a t 524 nm) is almost twice that shown for a chloroform solution of the fraction of Co((+)atc), assigned the A-trans structure by Chen and Everett.15 Since the change to a benzene solution is expected to increase the rotation by only ca. 6’30,14 this is a significant difference. It is clear that their fraction contained an appreciable amount of the cz isomer which has a generally opposite ORD curve from that of the P isomer. In the synthetic mixture, the /3 isomer is present in greater amounts than the CY isomer and has a stronger rotation (vide infra), and therefore Chen and Everett’s chromatographic fraction had an ORD curve with the general shape of that of the @ isomer. Likewise, their second fraction, assigned the A-cis structure, was actually a mixture of the y and 6 isomers with the y isomer in excess (vide infra). It should be noted here that observations similar to these can almost certainly be made on all the reports in the literature concerning the magnitude of the optical activity measured for the Co(hmc)s comIncomplete separation of all plexes .7a,b,11,13b, 14! 15,28, the isomers has resulted in incorrect assignments and conclusions. Nuclear magnetic resonance has been an invaluable aid in ensuring the purity and structural assignments of the isomers in the present study. The circular dichroism curves for the four isomers are shown in Figure 4. Some of the data are summarized 01
+532
-532
350
300
400
450 A
Figure 4.-The
500
550
600
lnrn)
circular dichrosim curves for the four isomers in benzene solution.
eo((f)atc)s
d9.0l
in
TABLE IV THECIRCULAR DICHROISM DATAFOR THE Foirrc Co(( +)atc)s ISOMERS~ ?, cm-1
a
16,800
p
1‘3,700 16,700 16,800
y
B [AI
111A)
Isomer
-83 3
6 300
350
450
400
500
550
600
Alnrni
Figure 3.-The optical rotatory dispersion curves for the four Co((+)atc)s isomers in benzene solution.
a
-1Al
-.+
lEa-(el
- ar)max
-5.64 +10.8 f7.39 -4.33
-‘AI + (LEb, LA);, cm-1 ( e l - erImax 22,400 +7.36 -11.1 22,400 22,100 -7.79 +6.03 22,600
-Charge s, cm-1 28,400 28,400 28,400 28,400
transfer(el
- cr)max 1-138 -129 -100
+I16
These data were obtained from benzene solutions.
mers while the /3 and y isomers are (- ). More importantly, the bands a t -615 nm are seen to exhibit negative Cotton effects in the CY and 6 isomers and positive Cotton effects in the P and y isomers. The Cotton
in Table IV. Although the maximum symmetry possible for any of the isomers is C3, the spectral assignments of the transitions responsible for the optical activity in Co((+)atc), have been made with reference to idealized D3 symmetry by Chen and Everett.I6 These are the same assignments made originally by Mason and
(25) R . A. Palmer, R. C. Fay, a n d T. S. Piper, Inovg. Chem., 8, 875 (1964). (2‘3) G. VJ. Everett and Y .T. Chen, J .Amev. Chem. Soc., 92, 508 (1970). (27) I . Hanazaki, F. Hanazaki, and S. Nagakura, J . Chem. P h y s . , 60, 265 (1969).
(28) J. H. Dunlop and R. D. Gillard, Advan. Inoug. Chem. Radiochem., 9, 211 (196‘3). (29) R . D. Gillard in “Physical Methods in Advanced Inorganic Chemis1968, p try,” H. A. 0. Hill and P. Day, Ed., Interscience, New York, N. Y., 197.
TRIS [ (+)-3-ACETYLCAMPHORATO]COBALT(III) coworkers for C o ( ( + ) h m ~ ) a land ~ , ~are ~ ~ also used here. Thus, the degeneracy of the 'TI, state in octahedral symmetry is partially removed to give lA%and 'E (designated lEa) states in D3. Mason, et al., reported transitions a t 700 and 610 nm with weakly negative and strongly positive CD peaks, respectively, for Co((+)h m ~ ) ~ . l ~They ~ b ~ cassign the peaks to the 'A1 --t 'Az and 1.4' + lEa excitations, respectively, on the basis of their experience that the Ea transition usually shows the greater rotational strength and in order to be consistent with the results of Piper, whose single-crystal polarized spectra of C ~ ( a c a cindicate )~ that in this compound the trigonal splitting constant, K , is positive (ie., the 'E, state lies a t higher energy than the lAz state).30 The C D peaks here reported at 600 nm are similarly assigned. Instrumental limitations prevented us from observing the 'A1 + 'A2 component. The lTZg state goes over into and 'E('Eb) states in D3symmetry. The 'A1 + lA1 transition is symmetry forbidden in D3,31so Mason and coworkers tentatively assign the C D peak that they observe a t 440 nm for C o ( ( + ) h m ~ to ) ~the 'A1 --t 1 E b t r a n s i t i ~ n . ~ Concern ~*~,~ has been expressed that the sign of rotation of this band (negative) is opposite that of the lower energy transition assigned to 'A1 --t 1Ea.2gHowever, it is clear from the theoretical treatment of Karipides and Piper that when K is positive, the signs of the Ea and Eb transitions will be opposite.32 Also, in the lower symmetry of these complexes, C3 or C1, the lA -+ lA transition would, of course, become allowed and could gain intensity through the action of a small component of the potential field serving to reduce the 0 3 symmetry. Thus, the assignment given in Table IV for the bands a t 445 nm is lA1 -+ (1Eb, lA). It is obvious that the sign of the Cotton effect of this band is opposite that of the lA1 + lEa transition for each isomer. The sign of the Cotton effect of the charge-transfer band a t 352 nm is always the same as that of the 'A + (lEb, 'A) band. Mason and coworkers have used the sign of the Cotton effect of the Ea band to assign the absolute configuration of Co((+)hmc), by analogy to their work with diaminecobalt(II1) complexes having five-membered chelate rings and Ne donor systems.13 Thus, a positive sign for this band was taken to indicate the A configuration. From similar considerations, Dunlop, Gillard, and Ug014 and Chen and Everett15 have assigned the A configuration to all the fractions of Co((+)hmc), or Co((+)atc), which they have isolated and the A configuration to the fractions of the Co((-)hmc), complex (Chen and Everett). However, the extension of this analogy to these systems with sixmembered chelate rings and oxygen donors is by no 32-34 and its use here will be means unequivocalZ6~ strictly tentative. Thus we assign the a isomer the A-trans structure, p the A-trans structure, y and 6 the A-cis and A-cis configurations, respectively. 35 The (30) T S Piper, J . Chem. Phys., 95, 1240 (1961). (31) W. Moffitt, ibzd., 25, 1189 (1956). (32) A G. Karipides and T S . Piper, i b i d , 40, 674 (1964), and references cited therein. (33) G. R . Brubaker a n d L. E . Webb, J . Amev. Chem SOC.,91, 7199 (1969). (34) R R . Judkins and D. J. Royer, Inovg Nucl. Chem. Left.,6,305 (1970). (35) These assignments have recently been confirmed by a n X-ray crystallographic determination of the absolute configuration of A-lvans-Cr((+)atc)a: W. D Horrocks, D. L. Johnson, and D. MacInnes, J . Anzev. Chem Soc., 92, 7620 (1970). King and Everett have shown by X-ray powder photography t h a t this complex is isomorphous with their B isomer (our p isomer).l'
Inorganic Chemistry, Vol. 10, No. 6,1971 1247 study of the circular dichroism of the exciton-coupled, long-axis-polarized T + T* transition of the p-keto enolate ligand bids to yield fruitful results in the determination of the absolute configuration of tris(P-keto enolate) complexes.36J7 Instrumental limitations have prevented us to date from investigating this transition which occurs a t 229 nm in Co(aca~)3.~'The energies of the T and T* levels in the 3-acylated camphorato ligands would be expected to be altered somewhat by the strain in the camphor ring system.16 The fact that the ligand itself is chiral would, of course, also have to be taken into account in a study of the CD of this T + T* excitation. In order to determine the equilibrium distribution of isomers, a solution of a!-Co((+)atc), in benzene was heated for 60 hr at 60" in a sealed nmr tube as described in the Experimental Section.as A reproduction of the spectrum obtained after such treatment is presented in Figure 5. In the spectrum shown, the acetyl methyl region was recorded a t a slightly greater spectrum amplitude setting than the 8-, 9-, and 10-methyl region.. However, in the spectra analyzed, both regions were recorded a t the same settings. The resolutions shown in the figure were done by manual analysis. Those in the crowded regions of the spectrum (-2.0, - 1.1, and -0.7 ppm) are only approximations. (Accurate resolutions in these regions were found to tax even an electronic curve resolver, such that inconsistent results were obtained.) Accurate results were obtained from the following well-resolved peaks: the a! peak a t -1.91 ppm, the y peaks a t -1.03 and -0.90 ppm, and the P peaks a t -0.97, -0.93, and -0.86. Complete resolutions of these peaks were made with the aid of the line widths known from the spectra of the pure isomers. The relative amounts of the a, 0, and y isomers were obtained by weighting the areas of these peaks by the number of methyl groups they represent. The values for the P and y isomers were obtained from the averages of their several peaks. There is no cleanly resolved peak for 6 and its relative abundance was determined by difference from the highly degenerate transitions a t -2.0, -1.1, and -0.7 ppm. The mole percentages of the isomers obtained by such treatment are as follows: a!, 31.3 f 0.87'; p, 45.2 f 1.3700; y, 16.4 f 1.0%; 6, 7.0 f 1.0%. These results were reproduced in separate experiments. Essentially identical results were obtained from similar analyses of the reaction mixture but the presence of impurity peaks limited the accuracy of the integrations of the isomer resonances.
Discussion Since the four stereoisomers of tris [ (+)-3-acetylcamphorato]cobalt(III) bear a diastereomeric relationship to one another, there is no necessity for their having the same stabilities. Consequently, all of the investigators who have studied such complexes over the years have commented on the strong stereo~electivity~~11J~-~~ or even total stereospecificity28 which they exhibit. The results obtained for C ~ ( ( + ) a t c )in~ this investigation call all of the preceding studies into serious ques(36) E. Larsen, S. F. Mason, and G. H. Searle, Acta Chem. Scand., 20, 191 (1966). (37) For a recent review of this approach see B. Bosnich, Accounts Chem. Res., 2 , 266 (1969). (38) T h e kinetics of this reaction and the analogous ones of the other isomers are under investigation: C. S. Springer, t o be submitted for publication.
C. S.SPRINGER, JR., R. E. SIEVERS, AND B. FEIBUSH
1248 Inorganic Chemistry, Vol. 10, No. 6,1971 co(dslc)3 in C6H6
acetyl methyl region
camphor m e t h y l region
n
Figure 5.-The acetyl and 8-,9-, and 10-(camphor)methyl regions of the 1H nmr spectrum of an equilibrium mixture of the stereoisomers of Co((+)atc)s in benzene. The two regions were recorded a t slightly different spectrum amplitude settings.
tion. After approximately 50 hr a t GO", a benzene solution of A-trans-Co( (+)-atc)s was converted into an equilibrium mixture of the A-trans, A-trans, A&, and A-cis isomers. Analysis of this mixture indicates that at 60" the equilibrium ratios of the concentration of the trans to the cis isomer for both the A (2.8 f 0.3) and the A (4.5 f O.G) cobalt epimers are close to the statistically expected value of 3. Thus, with regard to these two pairs, stereoselectivity is negligible. For the two pairs whose members bear epimeric relationships about the chiral center at the cobalt, K = A-trans/Atrans = 1.4 0.1 and K = A-&/A-cis = 2.3 f: 0.4 a t 60". These reflect A G O values of -0.22 0.07 kcal/ mol and -0.55 f 0.12 kcal/mol, respectively, which indicate only a very slight stereoselectivity in favor of the A configuration. In tris chelates of chiral bidentate ligands where the interligand interactions are dominated by the ring conformation (;.e,) the 1,2-diamine ligands) stereoselective effects are on the order of 1 kcal or greater.@ Thus in tris(@-keto enolate) complexes steric interactions of the ring substituents are reduced; possibly by the slight folding of the chelate rings about the 0-0 line.35 Inspection of molecular models reveals no extreme interactions between the various components of the Co((+)atc)s molecules and this is borne out by the experimental results. A free energy difference of approximately 0.25 kcal was also found between the A-trans and A-trans isomers of the tris(L-1eucinato)cobalt(II1) complex a t 117°.39 This is to be expected since chelate rings in a-amino acid complexes are much less puckered than those in 1,2-diamino complexes.28 Although they reported no evidence for A-cis-V-. atc at^)^, Chen and Everett found that the A-trans isomer was more stable than the A-trans isomer by approximately 0.7 kcal a t 29°.16 The A-trans:A-cis ratio was also found to be statistical. The A-trans:
*
*
(39) I?. G . Denning a n d T.S . Piper, I n o v g . Chem., 5 , 1056 (1966).
A-trans ratio of V ( ( + ) h m ~ )seemed ~ to be about the same as that of V((+)atc),. Estimates of the trans:& ratio of the ii and A epimers of V ( ( + ) h m ~ could )~ not be made. Everett and Chen have also recently studied the Co(111) and V(II1) complexes of some novel chiral @-keto enolate ligands produced by the formylation of (+)and (-)-carvone and (+)-pulegone.26 They also found that the stereoselectivity evidenced is slight (
